Life In 19x19

Quotation reference: http://www.lifein19x19.com/forum/viewto ... 22#p194822

hyperpape wrote:

RobertJasiek wrote:You ask for a line exactly where influence stones are distinguished from non-influence stones, i.e., a set of criteria exactly which are versus are not fulfilled in combination. I cannot provide this exactness of algorithmic quality. All I have is ca. a dozen principles with which the distinction can be made in most practical cases.

Could you teach this to others to a level that they can apply it and reach the same verdicts as you on novel positions, or have you?

As an extension of my joseki evaluation theory, the principles for identifying and counting influence stones in joseki-like (non-middle-game-fighting) positions are in a draft of the (probably next) book I am, coincidentally, currently writing. I think that the principles are easy to understand and so their teaching in written form will hopefully enable every reader to reach the same verdicts in almost all practically occurring local positions. (Counting influence stones for application to the middle game is very much easier on average. Counting them in josekis is the relatively demanding thing.) Since it is an unpublished draft, I have not taught the principles explicitly to anybody yet.

(And no, the book is not about josekis, except for the related chapter. It just happens that the topic fits well into the book. By topic, it could have been in Joseki 3 - Dictionary but, at that time, I did not have worked out the principles.)

Reply to: http://www.lifein19x19.com/forum/viewto ... 34#p194934

The point about having a model theory for current territory (and another model theory incorporating that for the special application of josekis) is not to pretend accuracy but to have a working model. The theory of determining current territory provides 14 principles for the nature of reduction sequences and quite a few further principles for further details and the context of application (see Positional Judgement 1 - Territory). What I have got from professionals about counting stopped helping me beyond 3 kyu because of its too great fuzziness. My own theory about current territory is useful for me at 5 dan and has enabled me to win more games, such as the one against Ali Jabarin 1p. That professionals get better results than me does not mean that their explanation of interpreting fuzziness would be as good as their play.

Quotation reference: http://www.lifein19x19.com/forum/viewto ... 31#p194931

John Fairbairn wrote:I feel sure it's been used in several other English texts

The only English books I recall have explained or tried to explain are
- a few of my books (especially Joseki 2, Positional Judgement 1)
- Positional Judgement (Cho Chikun) with a half-hearted explanation

Then there have been online discussions. Other English books have not explained it well, AFAIK.

I have to assume that you have only ever seen the method once

Before writing Joseki 2, yes. What you call mainstream in Japan has not made it into the ca. 2000 (for non-Japanese readers) easily available Japanese books in Europe or Japan I have seen.

the method of offsets

I can only guess what you mean. So what is the method, please?

you are confusing precision and accuracy

I have just checked a dictionary: these two words are confusing because their meanings overlap to some extent:)

The theory has to be proven to be accurate first. You have not done that yet. Nor have pros.

1) My theory and the theory / theories of pros are not the same.

2) I ignore the meta-discussion about the exact meaning of 'accurate'.

3) A theory of current territory cannot be accurate in the sense of predicting which intersections will be territory at the game end because the game is dynamic and, e.g., allows more than one correct perfect play sequence.

4) Therefore a theory of current territory during the opening or middle game always is a model of a perferably very good approximation of the amount of "safe" territory in a current position.

What does it mean to prove accuracy for such a model of approximation in (4)? Not the amount of territory is approximated because the amount itself is undefined. What is approximated is the process of transforming territory regions (and possibly moyos) into territories by simulating predictions of expected endgames due to assumptions about middle game preparations for them. In designing such a model, there is some freedom of choice among assumptions that can be made. So instead of proving accuracy per se, one must prove clarity and consistency of the chosen model's assumptions and details.

Since I am on the level of many related principles and have not reached algorithmic level yet, I cannot make a formal, mathematical (accurate, if you like) proof yet for the unambiguity and consistency of my chosen model's assumptions and details. AFAIK, the pros for their models have not even stated principles and clear assumptions yet. So my model is far more advanced towards clarity and consistency. In fact, I have written a whole book about stating and applying the principles and assumptions, for which the reviewers point out the, ugh, accuracy. Yes, it is not the accuracy of mathematical proof yet.

Quotation reference: http://www.lifein19x19.com/forum/viewto ... 32#p194932

Charles Matthews wrote: Being strong is apparently about being fuzzy about the right things.

The great complexity of the game makes it impossible to be exact about everything, so you are right. This, however, must not be an excuse a la John Fairbairn or DrStraw to strive for fuzziness. We expect pros to be correct about reading, endgame counting and maybe other things (but they do make mistakes despite the myth). We know that Japanese pros like to think hours about the opening to play it "correctly" (but Takagawa admitted that it was all about playing compromises because the correct moves cannot always be found within the limited time). We also see them playing huge middle game fights ending in death and resignation. Being fuzzy then means nothing more than not knowing who kills whom to win the game.

Thus fuzzy is not good per se. As I have explained before, there are applications where approximations are good enough. Applications other than the possibly accurate aspects of the game and the sheer complexity of some reading tasks. And for the remaining applications, you are probably right at least that good decision-making about what may be left fuzzy / approximate is part (not the whole) of playing strength. It may well be an under-estimated topic.

I have just checked a dictionary: these two words are confusing because their meanings overlap to some extent:)

I ignore the meta-discussion about the exact meaning of 'accurate'.

I think this sums up why it's so frustrating for people trying to help you. You seem not to want to learn but just want to bolster your own theory, and will ignore or twist anything to do that.

Thus fuzzy is not good per se.

Again, not all of this and the original thread was about you or your theory. An important strand was teachability. No pro has said fuzzy is good per se. They have often said it is good enough for amateurs, or good enough even for pros, who have to make compromises under time limits. The fuzzy way to count territories is a lot easier and so is of earlier and more direct benefit to amateurs, but beyond that it teaches players to look at the right things. Counting stones and working out ratios might teach you a lot about arithmetic but teaches you next to nothing about go, but using approximations teaches you (especially your subconscious) a lot about go, and so is much more useful in the long run. Imagine you are a general marching an army. You come to a forest blocking your way. You can choose to dive into the forest and have your men hack down the trees so that the heavy artillery can be pulled through, and maybe be too late and tired out for the battle. This is equivalent to counting. Or you can climb a nearby hill or send out scouts to see how big the forest is. If it looks (approximation) not too big, you can say, "Hey lads, let's go round it."

I'm now done with these threads.

John Fairbairn wrote:

I ignore the meta-discussion about the exact meaning of 'accurate'.

I think this sums up why it's so frustrating for people trying to help you.

I continue to ignore meta-discussion.

You seem not to want to learn but just want to bolster your own theory, and will ignore or twist anything to do that.

If you re-read my previous message,...

So what is the method, please?

...you need not launch another meta-discussion.

No pro has said fuzzy is good per se. They have often said it is good enough for amateurs, or good enough even for pros, who have to make compromises under time limits. The fuzzy way to count territories is a lot easier and so is of earlier and more direct benefit to amateurs, but beyond that it teaches players to look at the right things.

You, and the pros teaching in this manner, mix correct and false things.

It is correct that
- there are time limits requiring compromises in thinking,
- there are cases when rough counting is good enough,
- rough counting is a lot easier.

It is wrong that
- rough counting is always good enough (it is insufficient for a) close games or b) deciding between every two variations with resulting similar judgements),
- rough counting provides more direct benefit for amateurs (it provides only the illusion, because strategic applications from rough counting can be right or wrong and one does not know which),
- rough counting is the easiest way of making a judgement of a count (simply guessing who is ahead is much faster and easier than even rough counting, and with just a bit of practice successful with a similar frequency),
- it teaches players to look at the right things (a) If players do not count at all, they have more time to look at the "right", non-counting things and b) accurate counting enables players to look at the right things much more than rough counting. The "right", non-counting things are, as a learning concept, essentially independent from learning counting in positional judgement. Teachers should not use counting as a excuse for bad teaching of "right", non-counting things.)

Counting stones and working out ratios might teach you a lot about arithmetic but teaches you next to nothing about go,

Counting presumes basic knowledge of arithmetics: addition, subtraction, multiplication, devision. It is immaterial whether counting motivates improving one's awareness or speed of arithmetic calculations.

Counting stones and points and doing the right calculations with the counts belongs to the most important go skills. Not just basic counting of territorial positional judgement, but many other concepts and methods of counting, calculating and interpreting the counts and calculations are extremely important. The more I study and unnderstand pro games, the more very basic concepts of counts and calculations I discover that every professional gets right (even if he should achieve it subconsciously and not aware of the aspects explicitly) and almost all amateurs are not aware of because traditional go theory has neglected to express and teach those concepts.

For example, I see very many beginners making the same mistake because beginners often do not count to 1. They create 1 weak group because they are not aware that having 1 weak group is much worse than having 0 weak groups. Next, they create the 2nd weak group. Because they have not counted to 1 before, then they do not count to 2 to understand the most easily that they hurt their own position. And so they sometimes create 5 new weak groups within their 5 successive moves. Counting is essential! Even trivial counting is essential.

Your statement is one of the worst advices I have ever seen about go theory. You dislike numbers, but this is no reason to teach disliking numbers and counting.

but using approximations teaches you (especially your subconscious) a lot about go, and so is much more useful in the long run.

Approximations, unclarity and hidden knowledge have been extremely great hurdles for me to overcome when confronted with them in professional books and teaching.